Effect of Reductive Property ofActivated Carbon on Total OrganicHalogen AnalysisY A O L I , X I A N G R U Z H A N G , * A N DC H I I S H A N G

Department of Civil and Environmental Engineering, TheHong Kong University of Science and Technology, Clear WaterBay, Kowloon, Hong Kong SAR, China

Received October 08, 2009. Revised manuscript receivedJanuary 31, 2010. Accepted February 08, 2010.

Total organic halogen (TOX) is a collective parameter and atoxicity indicator for all the halogenated organic disinfectionbyproducts (DBPs) in a water sample. TOX can be measured withthe adsorption-pyrolysis method based on Standard Method5320B. This method involves concentration of organic halogensfrom water by adsorption onto activated carbon (AC) andremoval of inorganic halides present on the AC by competitivedisplacement by nitrate ions. Since AC can also act as areductant, this work studied whether the reduction of chlorinatedDBPs by AC occurs during the TOX measurement, to whatextent the reduction affects the measurement of TOX, what typeof chlorinated DBPs can be reduced by AC, and whether themethod for the TOX measurement can be improved. Initially,chlorinated Suwannee River fulvic acid samples were preparedand pretreated with precipitation/dialysis/ultrafiltration tominimize the chloride levels in the samples. It was found thatthe fractions of TOX in the precipitated, dialyzed, andultrafiltered samples that were reduced by AC in 5 min werearound 13%, 20% and 24%, respectively. The formation of someN-chloroamino compounds and their reactivity with AC wereexamined. The results indicate that organic chloramines are onetype of DBPs in TOX that could be reduced by AC. It wasdemonstrated that slight oxidation of AC with ozone basicallyinhibited its reduction for TOX and meanwhile maintainedits adsorption capacity for TOX.

Introduction

Natural organic matter (NOM) and bromide/iodide in rawwaters react with disinfectants to produce numerous halo-genated disinfection byproducts (DBPs), which have beenconsidered to be one major contributor to the human healthrisk of drinking water (1, 2). A collective parameter for all thehalogenated organic compounds is total organic halogen(TOX) (3). Since the discovery of haloforms in drinking waterin the 1970s (4), TOX as a significant parameter has beenstudied and applied in more than 800 journal papers(searched by SciFinder Scholar). These papers can be sortedinto five groups. Group I focuses on developing a TOXmethod. Jekel and Roberts summarized the early develop-ment of the TOX method, which entailed concentrating theorganics by adsorption, rinsing inorganic halide ions withnitrate, converting organic halides to hydrogen halides, and

detecting halides by microcoulometry (ref 5 and referencestherein). The development of the TOX method in recent yearslies in the differentiation of total organic chlorine (TOCl),total organic bromine, and total organic iodine by detectinghalides with off-line ion chromatography (6). Groups II, III,and IV focus on examining the formation of TOX underdifferent source water characteristics and disinfection con-ditions (7-9), the correlations between TOX and individualDBPs/NOM (10-13), and the associations between TOXexposure and toxicity (14), respectively. Group V focuses onexploring the gap between TOX and known specific DBPs(10, 15, 16). This gap has been the primary driving force infinding new halogenated DBPs (17-26). The parameter TOXperforms like “a master parameter”. The significance of TOXto DBP studies cannot be overestimated.

TOX can be measured with the adsorption-pyrolysismethod based on Standard Method 5320B (27). The first twosteps of this method involve concentration of organichalogens from water by adsorption onto activated carbon(AC) and removal of inorganic halides present on the AC bycompetitive displacement by nitrate ions. Besides as anadsorbent, AC can also act as a reductant. It has been reportedthat inorganic chloramines in drinking water can be reducedby AC to chloride (28). If some halogenated organic DBPswere reduced to inorganic halides when contacting with AC,they would be removed from the AC during rinsing withnitrate, leading to certain systematic error for the measuredTOX.

In view of the significance of the parameter TOX and thepossible error involved in the TOX measurement, the primaryobjective of this work was to examine whether the reductionof chlorinated DBPs by AC occurs during the TOX measure-ment and to what extent the reduction affects the measure-ment of TOX. In chlorinated NOM samples, the concentra-tions of Cl- are generally 8-12 times greater than those ofTOX; if partial TOX in a chlorinated NOM sample werereduced to Cl- by AC in the adsorption step, the Cl- incrementwould be insignificant compared to the Cl- level in the originalsample and thus it would be difficult to measure. Todemonstrate the reduction of TOX by AC, the Cl- concentra-tion in the original sample should be minimized to areasonable level by using appropriate pretreatment proce-dures. The second objective of this work was to inspect whattype of chlorinated DBPs in TOX can be reduced by AC. Sinceorganic chloramines are suspected to be one type ofchlorinated DBPs that can be reduced by AC, several aminocompounds were used in this work as model compounds inpreparing organic chloramines, which would be tested fortheir reactivity with AC. The third objective of this work wasto investigate whether the method for the TOX measurementcan be improved. Once the reductive property of AC wasdemonstrated to cause some systematic error in the TOXmeasurement with the standard method, it would be rationalto improve the method by inhibiting the reductive propertyof AC while its adsorption capacity is maintained. It wasexpected that slight oxidation of the AC by an oxidant mightinhibit its reductive property.

Experimental MethodsPreparation and Pretreatment of Chlorinated Samples. Allsolutions used were prepared with ultrapure water (18.2 MΩ/cm) supplied by a NANOpure system (Barnstead). A hy-pochlorite stock solution was prepared monthly by theabsorption of ultra-high-purity chlorine gas with a 1.0 MNaOH solution. Suwannee River fulvic acid (SRFA) from the

* Corresponding author phone:+852-2358-8479; fax:+852-2358-1534; e-mail: xiangru@ust.hk.

Environ. Sci. Technol. 2010, 44, 2105–2111

10.1021/es903077y 2010 American Chemical Society VOL. 44, NO. 6, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 2105

Published on Web 02/16/2010

International Humic Substances Society was dissolved intoultrapure water to prepare a SRFA stock solution.

Two chlorinated SRFA samples were prepared. For sampleI, the initial concentrations of SRFA and NaOCl were 3 mg/Las C and 5 mg/L as Cl2, respectively, which were used tosimulate the typical concentrations in drinking water treat-ment. For sample II, the initial concentrations of SRFA andNaOCl were 30 mg/L as C and 50 mg/L as Cl2, respectively,which are approximately 10 times higher than the typicallevels of NOM and Cl2 in drinking water treatment, to amplifythe possible reactions and products. The pH values of bothsamples were around 7.0. After a reaction time of 5 d underambient temperature, no free chlorine residuals weredetected in the samples with the DPD ferrous titrimetricmethod (29). Samples I and II had a volume of 10 and 2 L,respectively, and both were kept in a cold room at 4 °C priorto pretreatment.

Three pretreatment procedures including precipitation,dialysis, and ultrafiltration were used to minimize Cl- in thechlorinated SRFA samples. The pretreatment procedures aredetailed in the Supporting Information. Briefly, AgNO3 wasused to precipitate Cl-; dialysis was conducted in a semi-permeable membrane bag with a nominal molecular weightcutoff of 100 Da; ultrafiltration was conducted with a celluloseacetate membrane with a nominal molecular weight cutoffof 500 Da. Before and after the pretreatment, the concentra-tions of TOX and Cl- in each sample were measured.

Eight amino compounds were used as model compounds,including glycine, glutamate, leucine, glycylglycine, cytosine,adenine, phenylalanine, and tryptophan. For the preparationof a chlorinated model compound sample, the initialconcentrations of an amino compound and NaOCl were 1.0and 0.1 mM, respectively; the pH of the sample was around7.5. After a reaction time of 2 h under ambient temperature,no free chlorine residual was detected in the sample with theDPD ferrous titrimetric method (29).

Reactions of (Pretreated) Chlorinated Samples with AC.For either a pretreated chlorinated SRFA sample or achlorinated amino compound sample, four 20-mL aliquotsof the sample were collected into four vials. One aliquot wasused for blank control, and the other three aliquots wereallowed to react with AC. The AC was purchased fromMitsubishi (coconut-based with particle sizes of 100-200mesh and a very low halide background of e0.4 µg of Cl/40mg of AC), and was the same as the one packed in the ACcolumns for TOX analyses. Each of the three aliquots wasspiked with 40 mg of the AC and was adjusted to pH 2immediately with sulfuric acid (to simulate the TOX mea-surement procedure). After a contact time of 5 min, 12 h, or24 h, the aliquot was filtered with a syringe coupled with a0.45 µm Durapore PVDF membrane filter (Millipore Corp.),followed by rinsing the syringe filter with 10 mL of a 5000mg/L nitrate solution; the filtrate (30 mL in total) was collectedand adjusted back to pH 7 for determination of the Cl-

concentration. The syringe filter was then rinsed twice (torinse out all the Cl- in the AC and the syringe filter), eachtime with 20 mL ultrapure water and 10 mL of the nitratesolution; the Cl- concentration in each filtrate (30 mL intotal) was measured. Finally, the total Cl- concentration ineach aliquot after contact with the AC was calculated bysumming the Cl- concentrations in all the filtrates. For thealiquot used for blank control, it was treated in the same wayexcept that no AC was spiked, and thus it is designated asthe one with a contact time of “0 min” with the AC. Sincethe AC might contain some rinsable Cl- ions, another controlwas conducted as follows: 20 mL of ultrapure water was spikedwith 40 mg of the AC. After a contact time of 5 min or 12 hor 24 h, the sample was filtered with a syringe coupled witha 0.45 µm Durapore PVDF membrane filter, followed byrinsing the syringe filter with 10 mL of a 5000 mg/L nitrate

solution. The nitrate solution was found not to contain anymeasurable Cl- ions. The filtrate (30 mL in total) was collectedfor measuring the Cl- concentration with ion chromatog-raphy. The measured Cl- concentration would be deductedfrom the Cl- concentration in the aliquot with a contact timeof 5 min or 12 or 24 h with the AC.

To guarantee the success of the reduction experiment,the appropriate quality control is critical. All the glassware,syringes, and filters to be reused were intensively cleaned tominimize Cl- background. After cleaning, no Cl- was detectedin all the glassware and the syringes, and the Cl- blank fromthe filter or the 40 mg of AC was below 0.0005 mg. Theincrements of Cl- introduced from pH adjustment withdifferent acids (including sulfuric acid, nitric acid, andphosphoric acid) were compared. It was found that amongthe three acids tested, sulfuric acid introduced the leastamount of Cl-.

Detection of TOX, Cl-, and N-Chloroamino Compounds.TOX was measured by mainly following Standard Method5320B (27) except that an online ion chromatograph wasused as a halide detector. Briefly, a 40 mL aliquot of a samplewas passed through two consecutive prepacked AC columns(Mitsubishi) using a three-channel adsorption module(TXA03C, Mitsubishi). If the chloride content in the secondAC column was greater than 10% of the total value (first plussecond AC columns), the sample would be diluted. Then,the AC columns were washed with 10 mL of 5000 mg/L KNO3

as NO3- to remove inorganic halides and subsequently

subjected to pyrolysis at 1000 °C with an AQF-100 automaticquick furnace (Mitsubishi). The hydrogen halide and halogengases from the pyrolysis were absorbed in 5 mL of 0.003%H2O2 solution (freshly made daily and used for reducinghalogen gases to halide ions), which contained 2 mg/Lphosphate serving as an internal standard to estimate thevolume variations induced by the GA-100 gas absorptionunit (Mitsubishi). After absorption, an online ion chromato-graph (ICS-90, Dionex) was triggered and 200 µL of thesolution was injected. An IonPac analytical column (AS9-HC, Dionex) was used with a 9 mM Na2CO3 solution as theeluent at a flow rate of 1 mL/min. The concentrations of Cl-

were quantified by a conductivity detector. Since no bromideor iodide was involved in the sample preparation, the TOXlevels in the chlorinated samples were actually the TOClconcentrations. The practical quantitation limit for TOX was0.010 mg/L as Cl. The Cl- concentration in a sample wasmeasured with the same ion chromatograph under the sameinstrument settings. The relative standard deviations for theCl- measurement in seven aliquots of a standard NaClsolution (0.010 mg/L in ultrapure water) and a chlorinatedSRFA sample were 0.05% and 0.60%, respectively. Thepractical quantitation limit for Cl- measurement was 0.008mg/L. Unless otherwise specified, triplicates of a sample wereanalyzed for TOX and Cl-.

To demonstrate the existence of N-chloroamino com-pounds in chlorinated amino compound samples and toexamine the penetration of N-chloroamino compoundsthrough the AC, electrospray ionization tandem massspectrometry (ESI-MS/MS) analyses were conducted with aWaters Acquity triple quadrupole mass spectrometer (Wa-ters). The ESI-MS/MS analyses included full scan, precursorion scan of m/z 35 (24), product ion scan, and selected ionrecording (SIR) scan. The ESI-MS/MS parameters were setas follows: sample flow rate via an infusion pump, 10 µL/min; ESI negative mode; capillary voltage, 2.9 kV; cone voltage,15 V; source temperature, 110 °C; desolvation temperature,300 °C; desolvation gas, 650 L/h; cone gas, 50 L/h; LMresolution, 15; HM resolution, 15; collision energy, 15 eV;and collision gas (argon), 0.25 mL/min.

AC Treatment and Characterization. Our preliminaryresults showed that the AC treated with different chlorine

2106 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 6, 2010

doses could not meet the measurement requirement (i.e., toinhibit the reductive property of AC and meantime maintainits adsorption capacity), but the treatment with a specificozone dose worked well. In brief, ozone gas from an ozonegenerator (10K-2U, Enaly) was absorbed in 2 L of ultrapurewater. When the ozone concentration reached 2.4 mg/L,which was found to be the optimal ozone dose for treatingthe AC (Supporting Information, Figure S1), 10 mL of theozone solution was passed through an AC column im-mediately at a flow rate of 2 mL/min. The ozonated ACcolumn was kept in a fume hood for over 24 h until used forTOX analysis. The TOX recoveries for trichlorophenol withthe ozonated and original AC columns were tested.

The ozonated and original AC columns were then testedfor the TOX measurement on the basis of the standardmethod. Three samples were prepared as shown in theSupporting Information for the tests: chlorinated SRFA,ultrafiltered chlorinated SRFA, and chlorinated cytosine. Theultrafiltered chlorinated SRFA sample and the chlorinatedcytosine sample were also reacted with the ozonated andoriginal ACs to examine the Cl- increments.

To test the effect of quenching agents on the TOXmeasurement, aliquots of a chlorinated SRFA sample aftera chlorine contact time of 3 d (chlorine residual 0.80 mg/Las Cl2) or 5 d (no chlorine residual) were quenched withNaAsO2 or Na2SO3 for 0.5-1 h and then were measured forTOX with the original or ozonated AC columns.

To qualitatively determine the functional groups on theAC surface, some AC samples (the original AC, the ozonatedAC, and the ozonated AC that was immersed in ultrapurewater for 24 h) were analyzed by Fourier transform infraredspectroscopy (FTIR) using the PerkinElmer Spectrum BX.Each AC was dried in an oven at 100 °C for 12 h and thenmixed with KBr at an AC/KBr weight ratio of 1/300. Themixture was pressed at 8 ton for 5 min, followed by the FTIRanalysis at 64 scans and 4 cm-1 resolution. The elements inthe aforementioned AC samples were analyzed by X-rayphotoelectron spectroscopy (XPS) with the Physical Elec-tronics 5600 multitechnique system. The system was operatedunder a high vacuum of <10-9 Torr, using a pass energy of187.85 eV for surveys and 23.50 eV for chemical state withthe Kratos Analytical Axis Ultra instrument.

Results and DiscussionMinimization of Cl- Levels in Chlorinated SRFA Samples.Figure 1 shows the TOX and Cl- levels in chlorinated SRFAsamples before and after pretreatment. Detailed informationcan be found in the Supporting Information. In brief, duringthe pretreatment, Cl- was almost completely removed, butTOX was also partially removed, and the removal efficienciesof TOX varied among the pretreatment procedures (Figure1). In the precipitation, TOX was only slightly removed (18%).In the dialysis, TOX with nominal molecular weights lessthan 100 Da was removed (47%), while in the ultrafiltration,TOX with nominal molecular weights less than 500 Da wasremoved (83.2% of the TOX in the original sample passedthrough the membrane).

Reactions of Pretreated Chlorinated SRFA with AC. Thethree different types of pretreated chlorinated SRFA sampleswere used to examine the reactivities of TOX with differentMW distributions toward chemical reduction by AC. The5-min reaction with the AC was used to simulate theadsorption procedure in Standard Method 5320B (27). The12- and 24-h reactions were designed to examine if thereaction would continue after 5 min.

Figure 2a shows the Cl- concentrations in precipitatedsample I before and after reactions with the AC. The Cl-

concentration in precipitated sample I was 0.33 mg/L. Aftera contact time of 5 min and 12 and 24 h with the AC, the Cl-

concentration in the sample increased to 0.39, 0.47, and 0.56mg/L, respectively. The net Cl- increase in 5 min was 0.06mg/L, i.e., 0.06 mg/L of TOX was reduced to Cl- in 5 min.Since the measured TOX concentration in precipitatedsample I was 0.41 mg/L as Cl, the measurement error with

FIGURE 1. TOX and Cl-concentrations in the chlorinated SRFAsamples and in the samples treated with precipitation,ultrafiltration, and dialysis: (a) sample I, precipitated sample I,and ultrafiltered sample I and (b) sample II, 10 d dialyzedsample II, and 15 d dialyzed sample II.

FIGURE 2. Cl-concentrations in the pretreated chlorinated SRFAsamples after a contact time of 0, 5 min and 12 and 24 h withthe AC: (a) precipitated sample I and ultrafiltered sample I and(b) 10 d dialyzed sample II and 15 d dialyzed sample II.

VOL. 44, NO. 6, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 2107

the standard method can be estimated as 0.06/(0.41 + 0.06)) 12.8%. The results also indicate that more TOX in theprecipitated sample was reduced by the AC as the contacttime increased up to 24 h.

The Cl- concentration in 10-d dialyzed sample II was 1.72mg/L. After a contact time of 5 min and 12 and 24 h with theAC, the Cl- concentration in the sample increased to 2.15,2.44, and 2.64 mg/L, respectively (Figure 2b). The net Cl-

increase in 5 min was 0.43 mg/L. Since the measured TOXconcentration in 10-d dialyzed sample II was 1.55 mg/L asCl, the measurement error for the 10-d dialyzed sample withthe standard method is estimated to be 21.7%. Also basedon Figure 2b, the measurement error for the 15-d dialyzedsample with the standard method is estimated to be 19.4%.For either of the dialyzed samples, the fraction of TOX thatwas reduced by the AC increased significantly as the contacttime increased from 5 min to 12 h, while such a fraction didnot change much as the contact time increased from 12 to24 h.

Ultrafiltered sample I also reacted with the AC. After acontact time of 5 min and 12 and 24 h with the AC, the Cl-

concentration in the ultrafiltered sample increased from 0.18to 0.31, 0.32, and 0.33 mg/L, respectively (Figure 2a). The netCl- increase in 5 min was 0.13 mg/L. With the measured TOXconcentration in the ultrafiltered sample (0.42 mg/L as Cl),the measurement error for the ultrafiltered sample with thestandard method is estimated to be 23.6%. Interestingly, thefraction of TOX in the ultrafiltered sample that was reducedby the AC did not change much as the contact time increasedfrom 5 min to 24 h.

In summary, the fractions of TOX in the precipitated,dialyzed, and ultrafiltered samples that were reduced by theAC in 5 min were around 13%, 20%, and 24%, respectively.The TOX fraction with higher molecular weights appears tobe preferentially/readily reduced by AC.

Reactions of Chlorinated Amino Compounds with AC.Eight different amino compounds were chlorinated to testwhether N-chloroamino compounds can be formed, whetherthe preformed N-chloroamino compounds can react withAC, and whether the preformed N-chloroamino compoundscan penetrate through AC.

The DPD ferrous titrimetric method (29) was used to detectfree chlorine and combined chloramines as well as theexistence of organic chloramines. No residual free chlorinewas detected in all the chlorinated amino compound samples,but after the addition of some KI, three chlorinated aminocompound samples with glycine, glycylglycine, and cytosineturned to red, which means that three N-chloroaminocompounds might be formed, including N-chloroglycine,N-chloroglycylglycine, and N-chlorocytosine. Meantime,these chlorinated amino compound samples were allowedto react with the AC for 5 min; after reaction, no organicchloramines were detected in any of these samples. Thissuggests that the formed N-chloroamino compounds werereactive with the AC.

Figure 3 shows the ESI-MS full scan spectra of thechlorinated amino compound samples. For the chlorinatedglycine sample, the molecular ion cluster at m/z 108/110was observed, which should correspond to N-chloroglycine[HN(Cl)CH2COO-]. The intensity ratio of m/z 108/110 was

FIGURE 3. ESI-MS full scan spectra of different chlorinated amino compound samples: (a) glycine, (b) glycylglycine, (c) cytosine, (d)adenine, (e) glutamate, (f) leucine, (g) L-phenylalanine, and (h) L-tryptophan.

2108 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 6, 2010

deviated from 3:1 because the pair was partially overlappedwith another molecular ion cluster at m/z 110/112, whichcorresponded to the adduct of glycine with HCl[HCl ·H2NCH2COO-]. For the chlorinated glycylglycine andcytosine samples, the molecular ion clusters at m/z 165/167and 144/146 were observed, which should correspond toN-chloroglycylglycine [HN(Cl)CH2CONHCH2COO-] and N-chlorocytosine [HN(Cl)C3H2NCON-], respectively. For all theother chlorinated amino compound samples, no molecularion clusters were observed for the corresponding N-chlo-roamino compounds, although the adducts of amino com-pounds with HCl were present in all these samples.

As shown in the Supporting Information (Figure S2), theions with m/z 108, 165, and 144 were detected by the ESI-MS/MS precursor ion scans of m/z 35 of the chlorinatedglycine, glycylglycine, and cytosine samples, respectively.These ions corresponded to N-chloroglycine, N-chlorogly-cylglycine, and N-chlorocytosine, respectively. The Support-ing Information (Figure S2) also shows the ESI-MS/MSproduct ion scans of m/z 108, 165, and 144 of the chlorinatedglycine, glycylglycine, and cytosine samples. All the threemolecular ions generated a strong product ion at m/z 35(35Cl-).

Another relevant issue is whether N-chloroamino com-pounds can penetrate through AC. If they could completelypenetrate through AC, it would be meaningless to study theirreaction with AC. To test this, ESI-MS/MS SIR scans wereperformed to detect N-chloroamino compounds in thechlorinated amino compound samples before and after thesesamples passed through an AC column. According to theSupporting Information (Figure S3), the three N-chloroaminocompounds were detected at significant levels in thechlorinated amino compound samples, while only tiny peaksor no peaks corresponding to the N-chloroamino compoundswere observed in the samples passing through the AC column.It suggests that these N-chloroamino compounds basicallydid not penetrate through the AC.

The Supporting Information (Figure S4) shows the Cl-

concentrations in the chlorinated glycine, glycylglycine, andcytosine samples after reactions with the AC for 5 min and12 and 24 h. It indicates that the three N-chloroaminocompounds could be reduced by the AC, which provideddirect evidence for the reduction of partial TOX during contactwith AC.

Improvement of the TOX Method with Ozonated AC.The reductive property of the AC was demonstrated to causethe systematic error in the TOX measurement, so the ACcolumns were slightly oxidized by ozone. When the ozonatedand original AC columns were used in the TOX measurementprocedure, the recoveries for trichlorophenol were found tobe 94.8% and 91.8%, respectively. Trichlorophenol is thedesignated standard in the standard method for testing theTOX recovery and may not be reduced by the AC. The resultsindicate that slight oxidation of the AC with ozone still wellmaintained its adsorption capacity.

Then, the ozonated AC columns were used in the TOXmeasurement. As shown in Figure 4a, the TOX concentrationsin the chlorinated SRFA sample measured with the originaland ozonated AC columns were 0.48( 0.04 mg/L as Cl (from20 aliquot measurements) and 0.60 ( 0.02 mg/L as Cl (from20 aliquot measurements), respectively. The net TOX incre-ment was 0.12 mg/L as Cl. The TOX concentrations in theultrafiltered chlorinated SRFA sample with the original andozonated AC columns were 0.28( 0.00 and 0.35( 0.01 mg/Las Cl, respectively. The net TOX increment was 0.07 mg/L asCl. The TOX concentrations in the chlorinated cytosinesample with the original and ozonated AC columns were1.56 ( 0.14 and 1.97 ( 0.08 mg/L as Cl, respectively. The netTOX increment was 0.41 mg/L as Cl. The results suggest that

slight oxidation of the AC with ozone might effectively inhibitits reductive property.

To confirm the reduction inhibition with the ozonatedAC, two of the above samples were reacted with the originaland ozonated ACs. Figure 4b shows the Cl- concentrationsin the ultrafiltered chlorinated SRFA sample before and afterreactions with the ACs. After a contact time of 5 min, the Cl-

concentration in the sample with the original AC increasedto 0.43 mg/L, while that in the sample with the ozonated AC(0.35 mg/L) was close to the initial Cl- concentration (0.32mg/L). The Cl- decrement with the original and ozonatedACs was 0.08 mg/L, which was close to the correspondingTOX increment (0.07 mg/L as Cl). Figure 4c shows the Cl-

concentrations in the chlorinated cytosine sample beforeand after reactions with the ACs. After a contact time of 5min, the Cl- concentration in the sample with the originalAC increased to 3.48 mg/L, while that in the sample with theozonated AC (2.99 mg/L) was close to the initial Cl-

concentration (2.98 mg/L). The Cl- decrement with theoriginal and ozonated ACs was 0.49 mg/L, which was close

FIGURE 4. (a) TOX concentrations in different samplesmeasured with the original and ozonated ACs; (b)Cl-concentrations in ultrafiltered sample I after differentcontact time with the original and ozonated ACs; (c)Cl-concentrations in the chlorinated cytosine sample afterdifferent contact time with the original and ozonated ACs.

VOL. 44, NO. 6, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 2109

to the corresponding TOX increment (0.41 mg/L as Cl). Theresults demonstrated that the ozonated AC basically inhibitedthe reduction of TOX. Figure 4 also shows that the ozonatedAC, if it stayed longer in water, might recover the reductiveproperty of the original AC.

One concern is about the effect of quenching agents (suchas NaAsO2 and Na2SO3) on the TOX measurement with theoriginal or ozonated AC columns. The Supporting Informa-tion (Figure S5) shows that both quenching agents loweredthe TOX levels measured with either the original or ozonatedAC columns, but in the NaAsO2-quenched sample aliquots,the TOX level measured with the ozonated AC columns wassignificantly higher than that measured with the original ACcolumns.

Figure 5a shows the FTIR spectra of the original andozonated ACs. The peak at 1380 cm-1 may indicate carboxylor carbonated bonds (30) on the surface of the ozonated AC.The elemental compositions of the original and ozonatedACs detected by XPS are shown in Figure 5b. Compared tothe oxygen content in the original AC (1.45%), the oxygencontent in the ozonated AC (5.81%) increased significantly,suggesting a dramatic increase in the oxygen-containingfunctionalities on the surface of the ozonated AC. Thecoverage of -CdO, -COOH, or -COO- functionalities onthe ozonated AC surface may dramatically inhibit thereductive property of AC, but still effectively adsorb halo-genated DBPs (which are generally rich in -COOH, -OH,-NH2, or -CX functionalities) via hydrogen bonds. Figure5 also shows that if the ozonated AC was immersed in waterfor 24 h, the oxygen-containing functionalities on theozonated AC surface might be lost, resulting in the recoveryof the reductive property of the original AC. The proposedmechanisms for the adsorption and/or reduction of TOX on

the original and ozonated ACs are illustrated in the SupportingInformation (Figure S6).

AcknowledgmentsThe work was supported by a grant from the Research GrantsCouncil of the Hong Kong Special Administrative Region,China (Project No. HKUST622808). The authors thank GuoyuDing and Hongyan Zhai for their help in mass spectrometryanalyses. The authors are grateful to the Editor, threeanonymous referees, and Prof. Guohua Chen for their usefulsuggestions.

Supporting Information AvailableAdditional details and Figures S1-S6. This material isavailable free of charge via the Internet at http://pubs.acs.org.

Literature Cited(1) Richardson, S. D.; Plewa, M. J.; Wagner, E. D.; Schoeny, R.;

DeMarini, D. M. Occurrence, genotoxicity, and carcinogenicityof regulated and emerging disinfection by-products in drinkingwater: A review and roadmap for research. Mutat. Res. 2007,636, 178–242.

(2) Richardson, S. D. Water analysis: Emerging contaminants andcurrent issues. Anal. Chem. 2009, 81, 4645–4677.

(3) Dressman, R. C.; Stevens, A. A. The analysis of organohalidesin water-an evaluation update. J. Am. Water Works Assoc. 1983,75, 431–434.

(4) Rook, J. J. Formation of haloforms during chlorination of naturalwaters. Water Treat. Exam. 1974, 23, 234–243.

(5) Jekel, M. R.; Roberts, P. V. Total organic halogen as a parameterfor the characterization of reclaimed waters: Measurement,occurrence, formation, and removal. Environ. Sci. Technol. 1980,14, 970–975.

(6) Hua, G.; Reckhow, D. A. Determination of TOCl, TOBr and TOIin drinking water by pyrolysis and off-line ion chromatography.Anal. Bioanal. Chem. 2006, 384, 495–504.

(7) Hua, G.; Reckhow, D. A. DBP formation during chlorinationand chloramination: Effect of reaction time, pH, dosage, andtemperature. J. Am. Water Works Assoc. 2008, 100 (8), 82–95.

(8) Hua, G.; Reckhow, D. A.; Kim, J. Effect of bromide and iodideions on the formation and speciation of disinfection byproductsduring chlorination. Environ. Sci. Technol. 2006, 40, 3050–3056.

(9) Glaze, W. H.; Peyton, G. R.; Rawley, R. Total organic halogen aswater quality parameter: Adsorption/microcoulometric method.Environ. Sci. Technol. 1977, 11, 685–690.

(10) Singer, P. C.; Chang, S. D. Correlations between trihalomethanesand total organic halides formed during water treatment. J. Am.Water Works Assoc. 1989, 81 (8), 61–65.

(11) Yang, X.; Shang, C.; Lee, W.; Westerhoff, P.; Fan, C. Correlationsbetween organic matter properties and DBP formation duringchloramination. Water Res. 2008, 42, 2329–2339.

(12) Kristiana, I.; Gallarda, H.; Joll, C.; Croue, J.-P. The formation ofhalogen-specific TOX from chlorination and chloramination ofnatural organic matter isolates. Water Res. 2009, 43, 4177–4186.

(13) Li, C.; Benjamin, M. M.; Korshin, G. V. The relationship betweenTOX formation and spectral changes accompanying chlorinationof pre-concentrated or fractionated NOM. Water Res. 2002, 36,3265–3272.

(14) Savitz, D. A.; Singer, P. C.; Herring, A. H.; Hartmann, K. E.;Weinberg, H. S.; Makarushka, C. Exposure to drinking waterdisinfection by-products and pregnancy loss. Am. J. Epidemiol.2006, 164, 1043–1051.

(15) Krasner, S. W.; Symons, J. M. ; Speitel, G. E., Jr.; Diehl, A. C.;Hwang, C. J.; Xia, R.; Barrett, S. E. Effects of water qualityparameters on DBP formation during chloramination In Pro-ceedings of the AWWA Water Quality Technology Conference;Toronto, Ont., Canada, 1996; pp 601-628.

(16) Hua, G.; Reckhow, D. A. Comparison of disinfection byproductformation from chlorine and alternative disinfectants. WaterRes. 2007, 41, 1667–1678.

(17) Richardson, S. D.; Thruston, A. D., Jr.; Rav-Acha, C.; Groisman,L.; Popilevsky, I.; Juraev, O.; Glezer, V.; Mckague, A. B.; Plewa,M. J.; Wagner, E. D. Tribromopyrrole, brominated acids, andother disinfection byproducts produced by disinfection ofdrinking water rich in bromide. Environ. Sci. Technol. 2003, 37,3782–3793.

(18) Krasner, S. W.; Weinberg, H. S.; Richardson, S. D.; Pastor, S. J.;Chinn, R.; Sclimenti, M. J.; Onstad, G. D.; Thruston, A. D., Jr.Occurrence of a new generation of disinfection byproducts.

FIGURE 5. Characterization of the original AC, the ozonated AC,and the ozonated AC that was immersed in water for 24 h: (a)FTIR spectra and (b) XPS spectra.

2110 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 6, 2010

Environ. Sci. Technol. 2006, 40, 7175–7185.(19) Onstad, G. D.; Weinberg, H. S.; Krasner, S. W. Occurrence of

halogenated furanones in U.S. drinking waters. Environ. Sci.Technol. 2008, 42, 3341–3348.

(20) Zhang, X.; Talley, J. W; Boggess, B.; Ding, G.; Birdsell, D. Fastselective detection of polar brominated disinfection byproductsin drinking water using precursor ion scans. Environ. Sci.Technol. 2008, 42, 6598–6603.

(21) Zhai, H.; Zhang, X. A new method for differentiating adductsof common drinking water DBPs from higher molecular weightDBPs in electrospray ionization-mass spectrometry analysis.Water Res. 2009, 43, 2093–2100.

(22) Krasner, S. W. The formation and control of emerging disinfec-tion by-products of health concern. Philos. Trans. R. Soc. A 2009,367, 4077–4095.

(23) Zhang, X.; Minear, R. A. Formation, adsorption and separationof high molecular weight disinfection byproducts resulting fromchlorination of aquatic humic substances. Water Res. 2006, 40,221–230.

(24) Zhang, X.; Minear, R. A.; Barrett, S. E. Characterization of highmolecular weight disinfection byproducts from chlorination ofhumic substances with/without coagulation pretreatment using

UF-SEC-ESI/MS/MS. Environ. Sci. Technol. 2005, 39, 963–972.

(25) Hua, G.; Reckhow, D. A. Hydrophobicity and molecular sizedistribution of unknown TOX in drinking water. J. Environ. Eng.2008, 134, 152–160.

(26) Ding, G.; Zhang, X. A picture of polar iodinated disinfectionbyproducts in drinking water by (UPLC/)ESI-tqMS. Environ.Sci. Technol. 2009, 43, 9287–9293.

(27) APHA, AWWA, and WEF. Standard Methods for the Examinationof Water and Wastewater; Eaton, A. D., Clesceri, L. S., Greenberg,A. E., Eds.; APHA: Washington, DC, 1995; pp 522-527.

(28) Bauer, R. C.; Snoeyink, V. L. Reactions of chloramines with activecarbon. J. Water Pollut. Control Fed. 1973, 45, 2290–2301.

(29) APHA, AWWA, and WEF. Standard Methods for the Examinationof Water and Wastewater; Eaton, A. D., Clesceri, L. S., Greenberg,A. E., Eds.; APHA: Washington, DC, 1995; pp 443-445.

(30) Biniak, S.; Szymanski, G.; Siedlewski, J.; Swiatkoski, A. Thecharacterization of activated carbons with oxygen and nitrogensurface groups. Carbon 1997, 35, 1799–1810.

ES903077Y

VOL. 44, NO. 6, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 2111